I have been testing a L2TPv3 pseudowire setup for implementation as a layer 2 vpn solution over an Ethernet / IP backbone. My network hardware doesn't support MPLS so we have chosen to go with L2TPv3.
In the test environment I have managed to get the two Customer-Edge (CE) devices communicating with each other at layer 2 across the IP backbone.
As part of the testing process I ran a packet sniffer at different points in the CE and Provider (P) networks. From this I have discovered that L2TPv3-encapsulated packets are being spammed in one direction on the data channel without any apparent reason. This traffic appears to be generated at one of the Provider-Edge (PE) routers but I have been unable to determine the cause. Control channel messages are behaving normally.
The following image is a diagram of the test setup:
And the next shows the wireshark locations and numbers:
I have also attached the following in hopes that you might assist me in discovering the root cause of this issue:
- Wireshark feeds for each point in the path in both directions (in .pcap and .txt format)
- Configuration files for each device in the test case
Wireshark feed ws6 shows the spammed traffic (it is of the direction PE2 > PE1).
Please note that the Wireshark captures were taken over a 70 second period and in each case begun and concluded with 5 ICMP echo requests from a CE switch with datagram size set to 1400.
Happy to provide further information or alternative file formats.
Thanks in advance for your assistance.
Message was edited by: Michael Honeyman Moved to a different community. Also - the order of the spammed encapsulated traffic on the wire seems to go 98 bytes, 106 bytes, 102 bytes.
